The over-the-counter (OTC) derivatives market — with a gross market value of approximately 10 trillion USD — is the largest part of the financial trading market by volume. One of the recent developments in the OTC market is the introduction of initial margin (IM), which is a new collateral requirement for bilateral derivative trades, in addition to the existing variation margin. The cost of funding initial margin is called Margin Valuation Adjustment (MVA), and it forms a new component of the valuation adjustment framework for pricing derivative trades.
MVA requires the calculation of path-wise sensitivities of the portfolio which is computationally intensive. To deal with this computational challenge, we use a machine learning methodology, namely Gaussian process regression, to estimate the future sensitivity profiles. The second challenge in the context of IM is the optimization of MVA as a trading cost, due to its complex and non-linear nature. In this talk, I will present techniques to deal with both the computational and optimization aspect of MVA.
Sumit Sourabh is currently working as a front office trading Quant Analyst within Financial Markets at ING Bank in Amsterdam. In his role, he is responsible for the development and implementation of data-driven pricing and risk models for trading desks and risk departments.
Sumit has a joint position as a Research Scientist at the Informatics Institute, University of Amsterdam. He is associated with the European Union H2020 Bigdata Finance project on machine learning for trading and risk management. Prior to his current position, Sumit did a PhD in Mathematics funded by an EU Erasmus Mundus grant at the University of Amsterdam specializing in mathematical logic and theoretical computer science.